机车牵引变流器电容故障预测方法研究与应用

Research and Application of Capacitor Fault Prediction for Locomotive Traction Converter

为避免因牵引变流器电容器的性能退化而导致输入功率载荷加重,影响机车牵引系统的安全与可靠运行,本文提出一种电容器故障预测方法,其通过检测电容器输出电压纹波对电容参数进行辨识,并基于最小二乘支持向量机(LS-SVM)算法进行电容参数拟合来识别电容器退化特征,从而实现电容器故障预测。文章采用该方法与BP神经网络预测方法,以ESR为电容特征值为例,对牵引变流器中间直流回路的谐振电容器和支撑电容器进行故障预测。结果显示,LS-SVM模型误差小、精度高,更能反映实际变化;采用该LS-SVM模型对近2年现场电容器运行状态与故障情况进行预警验证,结果显示该方法预测准确率高于90%,验证了本文所提方法对电容故障预测的有效性。

In order to avoid the increase of input power load caused by the performance degradation of locomotive traction converter capacitor that affects the safe and reliable operation of a traction system,a capacitor fault prediction method is proposed in this paper.Capacitor parameters are identified by detecting output voltage ripples of capacitor,and capacitor parameters are fitted based on LS-SVM algorithm to identify the degradation characteristics of capacitor,so as to realize fault prediction of capacitor.Using this method and BP neural network prediction method,taking ESR as capacitance eigenvalue as an example,fault prediction of resonant capacitor and support capacitor in the middle DC circuit of traction converter is carried out.The results show that LS-SVM model has small error,high precision and can better reflect the actual changes.The LS-SVM model is used to pre-warning and verify the on-site operation states and fault situations of the capacitors in recent two years.The results show that the accuracy of this method is higher than 90%,which verifies the effectiveness of the proposed method for capacitor fault prediction.